September 28, 201312 yr So I need some help...disk red balled...stopped unraid...did smart reports no issues...checked cables....felt everything was O.K....so I unassigned drive then readded drive and started an array in hopes or rebuilding data....here is where I may have gone wrong...unraid reported that disk8 (the original red balled drive was unformated...so I checked the box to format it and allowed it to format the drive...don't know what I was thinking but the drive formated and the data never reconstructed...I quickly stopped the array, however the disk 8 reformatted...Any suggestions of how to proceed? Running unraid 4.7...syslog available on request...237K zipped so it exceeds the forum file size limit. Dan
September 29, 201312 yr Unfortunately it's too late at this point to rebuild it -- the parity info is no longer correct for that. For future reference, you have to Stop the array; unassign the drive; then START the array to show a missing drive; then Stop it gain; reassign the drive; and then Start the array -- at this point it will do the rebuild. NEVER allow a Format of a drive unless it's a new drive that you're adding to the array -- always get help for any other situation that asks you to format.
September 29, 201312 yr Author Gary, Thanks for the response...I did follow the procedure you outlined...however the rebuild did not start after reassigning the drive...what came up was that the drive was unformatted...unfortunately I assumed that the drive needed to be reformatted and that the rebuild would start at the end of the format.
September 30, 201312 yr Interesting that it didn't start the rebuild. Most of the time when this doesn't happen, it's because the drive was unassigned; then re-assigned; without an intervening Start and Stop of the array. The array MUST show the "Missing" drive when Started before you can Stop; Assign; Start and force the rebuild. Bottom line; I have no idea why it didn't work -- but since you formatted the drive it's too late to find out. If you haven't written anything to the drive yet, DON'T. There IS a chance that you could use Reiserfs to scan the drive for files and rebuild the directory tree. I am NOT a "Linux guy", but I know you can do a "rebuild tree" function that may let you recover some (probably not all) of your files. Personally, I think it's a MUCH better choice to simply recopy all of your data from your backups ... but I know not everyone has backups of their UnRAID data (they should !!). Search the forum for examples of the "rebuild tree" option -- or wait for a Linux guy to post to this thread ... or perhaps post a new question r.e. "How to do a Rebuild Tree".
September 30, 201312 yr If you have not written to the disk since it was "formatted", then there is a good chance that you could get most of the data back by putting the array into maintenance mode and running the command reiserfsck /dev/md?? --rebuild-tree --scan-whole-partition where ?? is the disk number from the unRAID GUI. This is likely to take some hours to run as it has to read the whole disk looking for files.
September 30, 201312 yr If you have not written to the disk since it was "formatted", then there is a good chance that you could get most of the data back by putting the array into maintenance mode and running the command reiserfsck /dev/md?? --rebuild-tree --scan-whole-partition where ?? is the disk number from the unRAID GUI. This is likely to take some hours to run as it has to read the whole disk looking for files. That's exactly what I thought the command was, but didn't want to post it, as I wasn't certain One note: the /dev/md?? is actually /dev/sd?? for SATA drives -- the first question mark is the associated letter (i.e. a, b, c etc. for sda, sdb, sdc, etc.); the second question mark is the partition number -- a 1 for all UnRAID disks. So the proper command for, say sdb, would be: reiserfsck /dev/sdb1 --rebuild-tree --scan-whole-partition
September 30, 201312 yr Author Thanks for the replies…they were just in time…I keep an offsite backup of my most important files (Photos, Home Videos and Documents) at work…my plan was to bring the array back online with disk 8 and copy over my offsite backup…I would have lost any data from the last month or so…but it wouldn’t have been the end of the world. So your suggestions are just in time as I have not written to the disk yet. So now I am running the reiserfsck as you have outlined above…it is in progress and it looks like it will be running most of the day. Now for some humor…disk 8 failed Saturday morning, we had a party at our house Saturday night and it was in the back of my mind all day…just didn’t have the time to work on it. At the party we decided to make Whole Fruit Margaritas in our new blender…after everyone left, I decided to sit down at the computer…a word to the wise don’t drink and repair unraid! There is some irony here…lime technology…margarita… Thanks again, Dan
September 30, 201312 yr If you have not written to the disk since it was "formatted", then there is a good chance that you could get most of the data back by putting the array into maintenance mode and running the command reiserfsck /dev/md?? --rebuild-tree --scan-whole-partition where ?? is the disk number from the unRAID GUI. This is likely to take some hours to run as it has to read the whole disk looking for files. That's exactly what I thought the command was, but didn't want to post it, as I wasn't certain One note: the /dev/md?? is actually /dev/sd?? for SATA drives -- the first question mark is the associated letter (i.e. a, b, c etc. for sda, sdb, sdc, etc.); the second question mark is the partition number -- a 1 for all UnRAID disks. So the proper command for, say sdb, would be: reiserfsck /dev/sdb1 --rebuild-tree --scan-whole-partition isnt /dev/md?? the unraid level, which you should use to ensure parity remains in sync? /deb/sdb1 would make parity out of sync, no?
September 30, 201312 yr isnt /dev/md?? the unraid level, which you should use to ensure parity remains in sync? /deb/sdb1 would make parity out of sync, no? You are correct. If you are operating on a functioning member of the parity protected array, virtual or real, you always want to use the md? device. If you are working with a drive attached to the unraid box but not part of the parity protected array, you use whatever base sd? or hd? device applies, and if you want to work with the first partition, you use sd?1 or hd?1. Any operations done on the sd? or hd? devices that are part of the parity protected array will break parity, and necessitate a correcting parity check to fix.
September 30, 201312 yr Any operations done on the sd? or hd? devices that are part of the parity protected array will break parity, and necessitate a correcting parity check to fix. /quote] Good to know. I wasn't aware of that difference -- especially since UnRAID shows them all as sdx devices. In this case it doesn't matter, since parity was already bad ... but clearly if that wasn't the case it would be nice to preserve parity while doing the tree rebuild.
October 1, 201312 yr Author So - I still may be out of luck....the process completed and this is what I got at the end...something about deleting 2987 unreachable items...while it was processing I saw a lot of recognizable file names, however when I go to /mnt/disk8 all I see is a mysql folder (which I did have installed originally on this disk with my XBMC library)...no other files. This is what I ran: reiserfsck /dev/sdk1 --rebuild-tree --scan-whole-partition Oddly enough...now the drive is flashing blue and says disabled disk present...at this point I am not going any further as I assume that if I rebuild the disk, it will rebuild it as a blank formatted drive. Any suggestions? Dan Flushing..finished Leaves inserted item by item 2449 Pass 3 (semantic): ####### Pass 3 ######### /mysql/MyVideos75/files.MYIvpf-10680: The file [57417 57465] has the wrong block count in the StatData /2012-04-24/IMG /SageTV//SageTV/Elementary-Heroine-5926454-0.mpgvpf-10680: The file [1255 3320] has the wrong block count in the StatData (12735640) - corrected to (1270.mpgvpf-10670: The file [1255 11323] has the wrong size in the StatData (3050727759) - corrected to (6108450816) vpf-10680: The file [1255 11323] has the wrong block count in the StatData (5958456) - corrected to (11930560) ... ackup/Bluefins Quickbooks Backup/capital (Backup Sep 11,2013 05 19 PM).QBBvpf-10680: The file [59709 59710] has the wrong block count in thFlushing..finished e [15404 67872] has the wrong block count/The Girl With The Drago Files found: 56284 Directories found: 4078 Symlinks found: 1691 Files with fixed size: 1 Pass 3a (looking for lost dir/files): ####### Pass 3a (lost+found pass) ######### Looking for lost directories: Looking for lost files:4 /sec rewrite_file: 2 items of file [1255 19521] moved to [1255 1012] vpf-10670: The file [1255 1012] has the wrong size in the StatData (0) - corrected to (3428352) vpf-10680: The file [1255 1012] has the wrong block count in the StatData (387586) - corrected to (6696) rewrite_file: 2 items of file [1255 47802] moved to [1255 1222] rewrite_file: 2 items of file [1255 56840] moved to [1255 1630] rewrite_file: 2 items of file [1255 57336] moved to [1255 1644] rewrite_file: 2 items of file [1255 58701] moved to [1255 3294] rewrite_file: 2 items of file [1255 58703] moved to [1255 3297] rewrite_file: 2 items of file [1255 58704] moved to [1255 3428] rewrite_file: 2 items of file [1255 58798] moved to [1255 3429] rewrite_file: 2 items of file [1255 58801] moved to [1255 5619] rewrite_file: 2 items of file [1255 58806] moved to [1255 8376] Flushing..finished Objects without names 10 Files linked to /lost+found 10 Objects having used objectids: 10 files fixed 10 Pass 4 - finished done 488330, 26 /sec Deleted unreachable items 2987 Flushing..finished Syncing..finished ########### reiserfsck finished at Mon Sep 30 20:35:11 2013 ###########
October 1, 201312 yr I'd write off the contents of that disk -- since the reiserfsck was done without maintaining parity you actually still have the parity you started with, so you COULD do a rebuild ... but as you noted it would simply be of the formatted disk, and a subsequent reiserfsck using the md designator would simply produce the same results (except with parity updated). I'd just remove that disk -- it was red-balled for a reason, so assume it's bad; do a New Config without it; and do an initial parity sync so you have good parity for the remaining disks. Then copy what you have available from your backups.
October 1, 201312 yr Author Is there any linux command to find out what files were on that drive...the reason, I ask is that during this process, I saw/recognized many files...but they just flashed on the screen. Is there some way to generate an output? Dan
October 1, 201312 yr I think it puts all the files it recovered in a folder with a name I don't recall, but something like "Recovered Files" or something like that. If you simply do a DIR of that disk (from a login prompt on the server ... i.e. DIR \dev\sdx (where x is the right letter) it should be fairly apparent -- then just do a DIR of that folder.
October 1, 201312 yr Most Windows-based file recovery utilities cannot work on the Reiserfs file system. You can, of course, install a Linux IFS that will let you access the UnRAID disk [the free LinuxReader works well for this: http://www.diskinternals.com/linux-reader/ ], but while this will let you access the disk from within Windows, it won't let Windows file recovery programs work on it.
October 1, 201312 yr By the way (for future reference), I found a note clarifying the mdx vs. sdx access ==> the UnRAID mdx device reference automatically refers to the first partition on the associated drive -- so you do NOT use a partition number when using that. In other words, if you're using the sdx reference, you must use the partition number (as I noted earlier) or you can corrupt the drive -- i.e. you have to use sdf1, but if you use mdf that automatically maps to the partition. Here's a quote from JoeL from another thread on this issue, with regards to the reiserfsck --rebuild-tree command: If you use reiserfsck on the raw drives you WILL cause the corruption. Only use reiserfsck on the FIRST PARTITION of the drives, or (much better) on the /dev/mdX devices (which unRAID ties to the first partitions on the disks)
October 1, 201312 yr By the way (for future reference), I found a note clarifying the mdx vs. sdx access ==> the UnRAID mdx device reference automatically refers to the first partition on the associated drive -- so you do NOT use a partition number when using that. In other words, if you're using the sdx reference, you must use the partition number (as I noted earlier) or you can corrupt the drive -- i.e. you have to use sdf1, but if you use mdf that automatically maps to the partition. Just a small point, using the md devices they are of the form /dev/md?? where ?? Is the disk number shown in the unRAID GUI (e.g. /dev/md12) rather than a letter as your mdf above suggests. That way the md device names are much easier to use and you do not need to be aware of the letter assigned to the sd? type device names.
October 1, 201312 yr According to JoeL you don't use a partition number with an mdX reference -- UnRAID automatically ties those references to the 1st partition on the drive. I suppose he could be wrong ... but that's RARE
October 2, 201312 yr According to JoeL you don't use a partition number with an mdX reference -- UnRAID automatically ties those references to the 1st partition on the drive. I suppose he could be wrong ... but that's RARE All itimpi was pointing out was that you used "mdf" in your post. The "md"'s are numeric so it would be md1, md2, md3, etc... not mda, mdb, ... mdf, ... like you were implying. The alphabetic designations are for the devices hd?1 or sd?1 (perferably). The rest of what you said was correct about unRAID already linking the md designation to the first partition (the "1" at the end of the hd?1 & sd?1).
October 2, 201312 yr Author As a follow up ...my unraid server is powered down...replacement disk arrived today...I hope to install tonight...should I preclear first (just anxious to start the disk replacement process). Dan
October 2, 201312 yr As a follow up ...my unraid server is powered down...replacement disk arrived today...I hope to install tonight...should I preclear first (just anxious to start the disk replacement process). Dan I would do a preclear just to confirm that the disk doesn't have any "infant mortality" issues.
October 2, 201312 yr Author I was afraid that was going to be the answer...I guess one more day won't kill me. Dan
Archived
This topic is now archived and is closed to further replies.